LFC

LFC estimates fold changes from RNA sequencing (RNA-seq) data by directly modeling count ratios to correct biases in high-throughput sequencing read counts.

Key Features:

• Count-ratio modeling: Directly models count ratios rather than relying on raw read counts to reduce bias in fold change estimation.
• Bias mitigation: Addresses biases that can affect more than 20% of genes called differentially regulated.
• Probabilistic framework: Employs a probabilistic model to provide a statistical foundation for estimates.
• Pseudo-counts theory: Provides a theoretical foundation for the use of pseudo-counts in fold change estimation.
• Credible intervals: Computes credible intervals around fold change estimates to quantify uncertainty.
• Normalization factors: Derives normalization factors intended to improve on those generated by existing methods.
• Validation with reference data: Performance has been compared against qPCR data from the MAQC/SEQC project and analyses of random barcoded sequencing data.

Scientific Applications:

• Differential expression analysis: Improves the accuracy and reliability of fold change estimation in RNA-seq differential expression studies.
• Gene regulation studies: Enhances interpretation of gene regulation across genomics and transcriptomics applications by reducing bias-driven misinterpretation.
• Benchmarking and validation: Serves in benchmarking and validation workflows using qPCR (MAQC/SEQC) and random barcoded sequencing data.

Methodology:

Direct modeling of count ratios using a probabilistic model, theoretical justification for pseudo-counts, computation of credible intervals, and derivation of normalization factors.